<empty/> element not processed

[lb42]

Issue TEIC/TEI#1596 has been resolved by creation of a new <empty/> element, but this is not currently being processed by any ODD processor.

[martindholmes]

@lb42 I'm looking at the example on the spec page:

http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-empty.html

and wondering what on earth it can mean. Why would anyone do that?

[martindholmes]

Also, is any processing actually required for this element? As far as we can tell, the right behaviour occurs already, just by doing nothing.

[sydb]

(“We” is me & Martin — tested <empty> as the only child of <content>, as the only child of <sequence>, and of <sequence> with one or two sibling <elementRef>s.)

[lb42]

@martindholmes It represents a content model which might be more economically expressed as (hi*). But as you can see in the discussion on TEIC/TEI#1596 economy of expression is not the primary goal here.

Can you confirm that for each pair of content models below, both expressions are processed identically?

(a) 

<content/>
 <content><empty/></content>

(b)

 <content>
       <elementRef key="hi" minOccurs="0" maxOccurs="1"/>
      </content> 

<content>
      <alternate>
          <empty/> 
           <elementRef key="hi"/>
       </alternate>
      </content> 

[martindholmes]

We think so (although our quick tests were slightly different), although might look at this ticket in a future Stylesheets group meeting and create a complete test suite for all the cases we can think of.

[sydb]

Council Stylesheets group did a few tests, and as far as we can tell all cases of <empty/> are being properly processed (that is, the output RELAX NG will validate the expected set of constructs) except for the case of an <empty> inside an <alternate>.
For example, @lb42’s example above:

<content>
  <alternate>
    <empty/>
    <elementRef key="hi"/>
  </alternate>
</content> 

results in just (hi). (It should result in either (hi?) or ( hi | empty ).)